DCT4—A New Member of the Dicarboxylate Transporter Family in C4 Grasses

Abstract Malate transport shuttles atmospheric carbon into the Calvin–Benson cycle during NADP-ME C4 photosynthesis. Previous characterizations of several plant dicarboxylate transporters (DCT) showed that they efficiently exchange malate across membranes. Here, we identify and characterize a previously unknown member of the DCT family, DCT4, in Sorghum bicolor. We show that SbDCT4 exchanges malate across membranes and its expression pattern is consistent with a role in malate transport during C4 photosynthesis. SbDCT4 is not syntenic to the characterized photosynthetic gene ZmDCT2, and an ortholog is not detectable in the maize reference genome. We found that the expression patterns of DCT family genes in the leaves of Zea mays, and S. bicolor varied by cell type. Our results suggest that subfunctionalization, of members of the DCT family, for the transport of malate into the bundle sheath plastids, occurred during the process of independent recurrent evolution of C4 photosynthesis in grasses of the PACMAD clade. We also show that this subfunctionalization is lineage independent. Our results challenge the dogma that key C4 genes must be orthologues of one another among C4 species, and shed new light on the evolution of C4 photosynthesis.


Introduction
Three subtypes of C 4 photosynthesis are generally recognized as defined by the primary decarboxylase in the bundle sheath (BS) cells: Chloroplastic NADP-dependent malic enzyme (NADP-ME); mitochondrial NAD-dependent malic enzyme (NAD-ME); and cytosolic phosphoenolpyruvate carboxykinase (PEPCK) (Hatch and Slack 1966;Hatch 1971;Rathnam and Edwards 1977). Different plant species may contain various Significance Dicarboxylate-transporter-2 (DCT2) plays a key role during C 4 photosynthesis in Zea mays. Its orthologs are assumed to function the same in related species, as Z. mays is the main C 4 reference species. We introduce a new gene, DCT4, that assumed the role of DCT2 in Sorghum bicolor and other C 4 grass species. By surveying related C 4 species, we propose that different members of the DCT family subfunctionalized for photosynthetic malate transport in the BS cells of C 4 grasses of the PACMAD clade. We suggest that rather than being static, biochemical adaptations continued after the divergence of the PACMAD lineages.
combinations of these three subtypes (Hatch 1971;Chapman and Hatch 1979;Furbank 2011;Pick et al. 2011;Wang, Brautigam, et al. 2014). The movement and exchange of malate across membranes, by dicarboxylate transporters (DCTs/ DiTs), plays a significant role during photosynthesis in NADP-ME and NAD-ME C 4 species (Ding et al. 2015). In C 3 plants, DCTs are crucial to nitrate assimilation, such as the GS/ GOGAT cycle and photorespiration (Linka and Weber 2010;Kinoshita et al. 2011). Taniguchi et al. characterized several plant DCTs that efficiently exchange malate across membranes (Taniguchi et al. 2002;2004). The differential expression of C 4 photosynthesis genes in mesophyll (M) and BS cells (John et al. 2014;Tausta et al. 2014;Wang, Czedik-Eysenberg, et al. 2014) suggests that different malate transporters may be needed to move malate out of the chloroplasts of M cells and into the chloroplasts of BS cells. In Zea mays, an NADP-ME C 4 grass, dicarboxylate transporter-2 (ZmDCT2, GRMZM2G086258) moves malate into the chloroplast of BS cells during C 4 photosynthesis (Weissmann et al. 2016). ZmDCT2 plays a critical role during C 4 photosynthesis in Z. mays, and its absence severely impairs plant growth and development (Weissmann et al. 2016). The role of DCTs in C 4 photosynthesis in other species, however, remains unknown.
Zea mays is the best characterized and functionally annotated C 4 grass species. As such, it is a useful reference for identification of photosynthesis-related genes in poorly characterized C 4 grasses and for resolving orthology (John et al. 2014;Ding et al. 2015;Huang et al. 2017). Microsynteny, the comparison of collinearity among related species, is a reliable approach to determine orthology and predict the function of a gene (Bennetzen and Freeling 1997;Chen et al. 1997;Tikhonov et al. 1999;Bennetzen 2000;Kumar et al. 2009;Jin et al. 2016). Davidson et al. (2012) showed that syntenic orthologs are likely to have conserved functions and expression patterns across lineages. Here, we identify a new member of the DCT family, DCT4, which is not syntenic to the photosynthetic gene ZmDCT2 and is not detected in the maize reference genome. We demonstrate that S. bicolor DCT4 (SbDCT4) efficiently exchanges malate across membranes, consistent with a malate transport role in C 4 photosynthesis. We characterize the diverse expression patterns of DCT genes in leaves of multiple grass species. We also propose that subfunctionalization of DCTs in grasses of the PACMAD clade (Sanchez-Ken and Clark 2010) occurred during independent recurrent evolution of C 4 photosynthesis.

Identification of DCT4 in S. bicolor
To learn more about C 4 -related DCT in species evolutionarily related to maize, we identified the syntenic ortholog of ZmDCT2 in S. bicolor. Two genes, Sobic.007G226700 and Sobic.007G226800, are present at the predicted syntenic orthologous position on chromosome 7. We refer to them as SbDCT2.1 and SbDCT2.2, respectively ( fig. 1). ZmDCT2 is abundantly expressed (table 1), and its expression is enriched in BS cells of maize leaves ( fig. 2) (Li et al. 2010;Tausta et al. 2014;Ding et al. 2015). In contrast, the expression profiles of SbDCT2.1 and SbDCT2.2 in S. bicolor leaves are low (table 1). SbDCT2.1 expression is slightly enriched in the M cells whereas SbDCT2.2 is enriched in BS cells ( fig. 2). We also analyzed the transcript levels of two other S. bicolor DCT, SbDCT1 (Sobic.002G233700) and SbOMT1 (Sobic.008G112300). These genes are the orthologs of the Z. mays genes ZmDCT1 (GRMZM2G040933) and Zm-oxoglutarate/malate transporter 1 (ZmOMT1; GRMZM2G383088), respectively. We found that SbDCT1 expression, similar to that of ZmDCT1, is relatively low (table 1), and only slightly differentially expressed in M cells relative to BS cells ( fig. 2). The expression of both ZmOMT1 and SbOMT1 is relatively high (table 1), and both are slightly enriched in M cells ( fig. 2).
In C 4 species, the expression of many photosynthetic genes is enriched in either BS and M cells (Li et al. 2010;John et al. 2014;Tausta et al. 2014;Weissmann et al. 2016;Rao et al. 2016). In NADP-ME species, two transporters, one within the BS cells and another in M cells, move malate in and out of the chloroplast during C 4 photosynthesis (Brautigam et al. 2008;Weissmann and Brutnell 2012;John et al. 2014;Tausta et al. 2014;Wang, Czedik-Eysenberg, et al. 2014). However, in sorghum leaves, we found only one highly expressed DCT, SbOMT1, that showed slightly enriched expression in M cells. Therefore, we screened the sorghum genome for additional homologs of known maize DCTs. We identified the gene Sobic.004G035500 that showed homology to ZmDCT1 and ZmDCT2 but was not syntenic to either gene ( fig. 1). DCT3 is the name of the second transcript of ZmDCT2 (Taniguchi et al. 2004), so we named this new gene SbDCT4. No syntenic ortholog of SbDCT4 is present in the reference genomes of Z. mays or Oryza sativa. The absence of syntenic conservation between S. bicolor and Z. mays and the lack of direct orthologs in Z. mays or C 3 species prevented identification of DCT4 in a previous bioinformatic screen for C 4 photosynthesis genes (Huang et al. 2017). The expression of SbDCT4 is moderately abundant (table 1)

SbDCT4 is an Efficient Malate Transporter
To verify the ability of SbDCT4 to transport malate, we cloned coding sequences from the three sorghum DCT genes, SbDCT1, SbDCT2, and SbDCT4. We measured the malate transport activities of the recombinant proteins expressed in yeast. SbDCT4 was an efficient malate transporter (table 2). The K m of SbDCT4 was similar to that of SbDCT1, and the affinity for malate was highest in SbDCT2 among the three SbDCTs (table 2), consistent with the relative malate transport activities reported for maize DCT1 and DCT2 (Taniguchi et al. 2004).

Phylogenetic Distribution of DCT Genes in Grasses
To understand the relationship of SbDCT4 to other grass DCT genes, we searched the genomes of the grass species Setaria italica, Urochloa fusca, Brachypodium distachyon, and Dichanthelium oligosanthes. In S. italica, an NADP-ME C 4 species, we identified a DCT, Seita.9G375100, that showed no syntenic orthologous relationship with DCT genes in other available grass genomes. Phylogenetic analysis showed that this gene clustered with SbDCT4 but not with SbDCT1 and Note that interspecies comparison is not possible, because expression levels were normalized within each species. SbDCT2 ( fig. 3). We designated this gene SiDCT4. We did not detect orthologs, syntenic or otherwise, in U. fusca, a PEPCK C 4 species, or in the two C 3 species. To expand the search for DCT4 in other grasses currently lacking genome assemblies, we examined leaf-derived transcript assemblies for Aristida congesta, Eriachne aristidea, Chasmanthium laxum, Danthoniopsis dinteri, Anthephora pubensis, Echinochloa esculenta, Paspalum vaginatum, and Arundinella hirta (Huang P, Mayfield-Jones D, Schnable J, Brutnell T, manuscript in preparation). We then used the predicted coding sequences of the DCT genes from available genomes and from the de novo leaf transcriptome assemblies to generate a phylogenetic tree of the DCT family. The resulting phylogeny shows that DCT4 transcripts form a distinct subclade from the DCT1 clade ( fig. 3). The absence of DCT4 transcript expression does not rule out the existence of the gene in the genome. We also used polymerase chain reaction (PCR) to survey for DCT genes in the genomes of grass species for which wholegenome assemblies were not available. We designed conserved primers (nondegenerate or minimally degenerate) to small regions unique to each of the three DCT genes using PrimaClade

Expression of Malate Transporter Genes in NADP-ME C 4 Grasses
C 3 species and U. fusca, a PEPCK C 4 species, express both DCT1 and DCT2 at low levels in leaves (table 1). C 4 NADP-ME species of the PACMAD clade generally express one DCT gene in leaves at a high level and also express one or two other DCT genes at low levels (table 1). We did not find an apparent lineage-specific pattern for the expression of the predominant DCT gene in the NADP-ME species we analyzed. This finding is consistent with random evolutionary processes underlying the subfunctionalization of members of the DCT family. Interestingly, Z. mays is the only species we examined in which DCT2 is the predominantly expressed DCT gene (table 1). We also examined the expression of the non-DCT malate transporter OMT1 gene in the leaves of grasses (table 1). Interestingly, we found that although OMT1 expression was generally abundant, there was no consistent pattern of relative expression between the DCT and OMT genes within the NADP-ME C 4 species (table 1).

Evolution of the DCT Gene Family in Grasses
We identified DCT4 as a new member of the DCT gene family in the grasses ( fig. 1). Our analysis suggests that DCT4 is present in some C 4 NADP-ME PACMAD grasses. DCT1 and DCT2 appear to have originated from a duplication of a single DCT gene after the monocot-eudicot split (Taniguchi et al. 2004) and DCT4 arose from a duplication of DCT1 at the root of the PACMAD grasses ( fig. 3). The expression of DCT genes in the grasses that we analyzed exhibited no clear lineage-specific patterns (table 1). Therefore, we propose that different members of the DCT family subfunctionalized for photosynthetic malate transport in the BS cells of C 4 grasses of the PACMAD clade.   Kinetic values from a previous report (Taniguchi et al. 2004).
This work also challenges the dogma that key C 4 genes must be orthologues of one another, among species, and show that they can be paralogs. This confirms the importance of including syntenic and expression data in assigning orthology across species, and of developing multiple models for C 4 photosynthesis in the grasses. For example, SiDCT4 was previously misannotated as the ortholog of ZmDCT2 (John et al. 2014), likely because of the lower expression level of SiDCT2 (Seita.9G375100) in leaf tissue. The use of different malate transporters, for example, DCT4 in S. bicolor and S. italica, or DCT2 in Z. mays, suggests that multiple evolutionary paths resulted in the development of an active C 4 NADP-ME photosynthetic cycle. It is interesting to note that common origins of C 4 photosynthesis are often defined based on the predominant decarboxylase utilized, thus maize and sorghum are considered to have evolved from a common C 4 ancestor. This analysis suggests that rather than being static, biochemical adaptations continued after the divergence of maize and sorghum lineages. Thus, optimizations of C 4 activities may be continuous as breeding pressures or climate change alters ecological niches of individual species.

Various C 4 Subtype Combinations Have Different Transport Requirements
The variation of expression levels among the different malate transporters within each NADP-ME species (table 1) suggests different transport requirements during C 4 photosynthesis. This supposition is in agreement with the view that the three subtypes of C 4 photosynthesis are mixed rather than exclusive (Hatch 1971;Chapman and Hatch 1979;Furbank 2011;Pick et al. 2011;Wang, Brautigam, et al. 2014). For example, Z. mays utilizes both the NADP-ME (75%) and PEPCK (25%) pathways to fix carbon (Chapman and Hatch 1979;Wingler et al. 1999;Weissmann et al. 2016), and has similar expression levels of DCT2 and OMT1 and low expression of DCT1. Sorghum bicolor moves carbon through both malate and aspartate, although no PEPCK activity was detected in its leaves (Chapman and Hatch 1979). Sorghum bicolor has similar expression levels for DCT4 and DCT1 and high expression of OMT1 (table 1). Other grass species may have DCT expression ratios that correspond to their unique combination of C 4 subsystems. For example, OMT1 is highly expressed in U. fusca, $3to 7-fold higher than DCT2 or DCT1, respectively. OMT1 transports dicarboxylates, excluding those containing an amino group (Taniguchi et al. 2002(Taniguchi et al. , 2004. Thus, in PEPCK C 4 plants, OMT1 may move oxaloacetate into the mesophyll chloroplast, and 2-oxoglutarate out, to support the high production of aspartate needed to maintain the photosynthetic cycle (Rathnam and Edwards 1977). Interestingly, both ZmOMT1 and SbOMT1 are only slightly differentially expressed in the M cells ( fig. 2). As the loss of DCT2 in Z. mays prevents movement of malate into the BS chloroplast (Weissmann et al. 2016), OMT1 cannot be moving malate into the BS chloroplast alongside DCT2. But OMT1 may also have a role in organic acid metabolism in both cell types, such as shuttling reducing equivalents in organelles other than the chloroplast (Pleite et al. 2005).

Conclusions
Our results show that the newly identified member of the DCT gene family, SbDCT4, is an efficient malate transporter. Based on the expression patterns of malate transporters among the grasses, we suggest that different members of the DCT family may have evolved multiple roles in C 4 photosynthesis. Further studies will be needed to verify the subcellular localization of these proteins and to define their specific metabolic functions. Characterizing the various combinations of C 4 photosynthetic subsystems in grasses will facilitate the exploitation of DCT genes, through breeding or engineering, to improve the performance of crop plants and increase yield.

Identification of DCT4 Genes in Sorghum and Other Grasses
We used QUOTA-ALIGN (Tang et al. 2011) to identify syntenic orthologous regions in grass species with sequenced genomes, following the protocol described in Zhang et al. (2017). To find homologous genes at nonsyntenic locations, we used two complementary approaches. For species with sequenced genomes, we used LASTZ (Harris 2007) to align the coding sequence of the primary transcript annotated in Phytozome (https://phytozome-next.jgi.doe.gov, last accessed January 6, 2021) to the genome assembly. For species without assembled genomes, we used LASTZ to align the coding sequence of the primary transcript from Phytozome to transcript assemblies generated by Trinity (Grabherr et al. 2011).

Measurements of Malate Transport
We cloned each of the three SbDCT cDNAs between the promoter and terminator of yeast GAL2 in the pTV3e vector (Nishizawa et al. 1995). We transformed the plasmids into yeast LBY416 cells and selected transformants on tryptophan-deficient agar plates. We prepared a crude membrane fraction from the selected yeast transformants. We used a freeze-thaw technique to reconstitute liposomes for the measurement of the uptake of [ 14 C]malate (Taniguchi et al. 2002).

Phylogenetic Analysis of DCT Homologs
DCT coding sequences for Z. mays, S. bicolor, S. italica, B. distachyon, O. sativa, D. oligosanthes, and U. fusca were from Phytozome (https://phytozome-next.jgi.doe.gov, last accessed January 6, 2021). We used BlastN (Altschul et al. 1990) to search de novo assembled leaf transcriptomes (Huang P, Mayfield-Jones D, Schnable J, Brutnell T, manuscript in preparation) from the C 4 grass species A. congesta, E. aristidea, C. laxum, D. dinteri, A. pubensis, E. esculenta, P. vaginatum, and A. hirta with the DCT sequences from maize, Setaria, and Sorghum as queries. We used ProGraphMSA to generate a codon-based sequence alignment (Szalkowski 2012). We used MEGA6 (Tamura et al. 2013), with default parameters and the branch support values based on 1,000 bootstraps, to generate the phylogenetic reconstruction with the maximum likelihood method and based on the nucleotides in the third position of codons (Simmons et al. 2006).

Analysis of Gene Expression for Decarboxylase Transporters in Grasses
For species with published leaf transcriptome profiles (Ouyang et al. 2007;Li et al. 2010;Zhang et al. 2012;Schnable 2014;Wang, Czedik-Eysenberg, et al. 2014;Studer et al. 2016), gene expression levels were calculated and normalized, for each species, as Transcripts Per Million (TPM). For the other species, the normalized TPM values were based on de novo transcriptome assemblies (Huang P, Mayfield-Jones D, Schnable J, Brutnell T, manuscript in preparation). The values in table 1 only allow for intraspecies comparisons among the decarboxylase transporters.